228 ABBOT: SOLAR RADIATION MEASUREMENTS 



changes are due to what we might call solar cloudiness, the 

 effects should not occur coincidentally in different directions 

 from the sun. 



One of the most convincing proofs of the essential soundness 

 of the measurements which indicate solar variability comes from 

 Clayton's investigations of terrestrial temperatures. These have 

 been published in Argentina, but also simultaneously by the 

 Smithsonian Institution in its Miscellaneous Collections. ^ 



In his first paper Clayton discusses the departures from nor- 

 mal temperatures for about 30 stations widely distributed over 

 the earth, as related to the "solar constant" values determined 

 on Mount Wilson. He does this largely by the mathematical 

 method of correlations, but not wholly so. 



If two quantities vary in the same sense wholly dependently 

 the one on the other, as for instance the lengths of the radii and 

 circumferences of circles, they are said to have a correlation 

 coefficient of +1. If the dependence is complete but the varia- 

 tions occur in opposite senses, as the widths and lengths of 

 rectangles of constant area, the correlation coefficient is — i . 

 Between these limits there are all magnitudes depending on 

 degrees of dependency between the two variables. If entirely 

 unrelated, the coefficient is zero. In this method of investiga- 

 tion, quite common in agriculture, eugenics and other sciences, 

 we must take account especially of the algebraic sign of the 

 coefficients and of their magnitude, compared to their probable 

 errors, in forming an opinion of the nature and degree of depend- 

 ence of the quaUties examined. 



As has been said, the Mount Wilson solar observations indi- 

 cated irregularly recurring variations sometimes reaching extremes 

 of 10 per cent in the solar heat available to warm the earth. 

 These changes often ran their course in a week or ten days. 

 Ranges of 10 per cent are rare but those of 2 or 3 per cent are 

 common. 



Clayton found that coefficients of correlation ranging from 

 +0.54 to —0.50 occtured as between solar and temperature 

 changes. His studies covered not only the day of the solar 



2 Vol. 68, No. 3, and Vol. 71, No. 3. 



